Polymer 112 (2017) 61e70

Contents lists available at ScienceDirect

Polymer
journal homepage: www.elsevier.com/locate/polymer

High-strength silk fibroin scaffolds with anisotropic mechanical

properties
Berkant Yetiskin, Oguz Okay*
Istanbul Technical University, Department of Chemistry, 34469 Istanbul, Turkey

a r t i c l e i n f o a b s t r a c t

Article history: In contrast to isotropic morphologies of synthetic hydrogels, many biological tissues possess anisotropic
Received 15 December 2016 hierarchical morphologies leading to extraordinary mechanical properties that cannot be mimicked by
Received in revised form synthetic materials. Here, we report preparation of anisotropic silk fibroin cryogels and scaffolds
20 January 2017
exhibiting a Young's modulus in the range of MPa that sustain up to 20 MPa compressive stresses. The
Accepted 28 January 2017
cryogels were prepared by a combined directional freezing e cryogelation process starting from an
Available online 31 January 2017
aqueous 4.2 wt% fibroin solution containing butanediol diglycidyl ether cross-linker and N,N,N0 ,N0 -tet-
ramethylethylenediamine. In the first step, the reactor containing the aqueous solution of fibroin, cross-
Keywords:
Silk fibroin
linker, and TEMED was immersed into liquid nitrogen at a controlled rate to create a directionally frozen
Anisotropic properties ice template. In the second step, cryogelation reactions were conducted in this frozen solution at 18  C
Cryogels whereby the cryo-concentrated fibroin in the unfrozen microzones of the reaction system forms a 3D
Scaffolds fibroin network. The scaffolds exhibit anisotropic microstructure and hence anisotropic mechanical
Macroporous structure properties, e.g., the Young's modulus is 3.4 ± 0.5 MPa and 0.8 ± 0.3 MPa when measured along the
directions parallel and vertical to the freezing direction, respectively. All the cryogels could completely be
compressed due to squeezing out of water from their pores. Upon removal of the load, the compressed
cryogels immediately recover their original dimensions and mechanical properties by absorbing the
released water into their pores.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction To prepare gels with anisotropic properties, several strategies

Hydrogels are chemically or physically cross-linked polymers
absorbing large quantities of water without dissolving [1]. Softness,
smartness, high water sorption capacity, and similarity to biological
tissues make hydrogels unique soft materials. However, there is still
a clear distinction between synthetic hydrogels and biological tis-
sues with regard to the microstructure and structure-related me-
chanical properties. In contrast to the isotropic morphologies of
synthetic hydrogels, many tissues including muscles [2], tendon [3],
cartilage [4], intervertebral disc [5], and cornea [6] possess aniso-
tropic hierarchical morphologies leading to extraordinary me-
chanical properties that cannot be mimicked by synthetic
materials. Biocompatible anisotropic hydrogels with a good me-
chanical performance have a broad range of potential applications
in tissue engineering, bioseparation, microfluidics, and organic
electronics [7].
* Corresponding author.
E-mail address: okayo@itu.edu.tr (O. Okay).
have been presented in the past years including directional freezing
[7,8], strain-induced reorientation [9e14], self-assembly [15e17],
dielectrophoresis [18], micropatterning [19], and 3D printing [20].
Directional freezing is a simple and promising approach to the
preparation of aligned porous materials [7,8]. By this technique, the
growth of solvent crystals during freezing of a polymer solution is
controlled in one direction, e.g., by immersing the reactor con-
taining the solution into a cold bath at a controlled rate. Uniaxial
freezing of the solutions of synthetic and natural polymers such as
polyvinyl alcohol (PVA) [21], agar [22], agarose [23], chitosan [24],
alginate [25,26], collagen [27], gelatin [28], soy proteins [29], silk
fibroin [30,31], or colloidal solutions of polymers and nanoparticles
[8], such as PVA and silica in a cold bath followed by removing
oriented solvent crystals via freeze-drying produce aligned porous
materials with various microstructures. However, such materials
generally dissolve in good solvents and exhibit poor mechanical
properties due to the absence of chemical cross-links inter-
connecting the polymer chains or the particles, limiting their
application areas. For instance, anisotropic scaffolds based on

http://dx.doi.org/10.1016/j.polymer.2017.01.079
0032-3861/© 2017 Elsevier Ltd. All rights reserved.
62 B. Yetiskin, O. Okay / Polymer 112 (2017) 61e70
chitosan/alginate produced by directional freezing exhibit a
Young's modulus of around 5 kPa [26], while gelatin scaffolds
rupture under 0.23 MPa compressive stresses [28]. However,
anisotropic scaffolds suitable for tissue engineering applications
should sustain high stresses in the order of MPa to protect their
integrity.
An alternative strategy is to combine the directional freezing
with the cryogelation technique to prepare mechanically strong
anisotropic gels. Cryogelation is a simple route to produce 3D
highly porous polymer networks of high toughness and superfast
responsivity [32,33]. Cryogelation reactions are conducted below
the freezing point of the reaction solution, during which the solvent
crystals are in equilibrium with the unfrozen liquid microchannels
containing cryo-concentrated reactants [32]. Thus, the reactions
only proceed in the unfrozen microchannels leading to the for-
mation of macroporous gels with thick pore walls. The combined
directional freezing e cryogelation approach was first reported by
our group in 2008 to produce butyl rubber (BR) organogels with an
aligned porous structure [34]. BR solution in cyclohexane was first
directionally frozen below the melting point of the solution and
then BR chains are intermolecularly cross-linked in the frozen so-
lution using sulfur monochloride as a cross-linker. The anisotropic
BR organogels are very tough and can be compressed up to about
99% strain without any crack development [34]. Several research
groups have recently reported the combination of the directional
freezing technique with redox-, gamma-, or UV-initiated cross-
linking cryopolymerization to prepare hydrogels based on chemi-
cally cross-linked poly (meth)acrylates with anisotropic morphol-
ogies and mechanical properties [35e40]. For instance, poly
(ethylene glycol) diacrylate scaffolds produced by directional
freezing e cryopolymerization technique exhibit a Young's
modulus of 8 kPa and 80 kPa during compressions along the di-
rections perpendicular and parallel to the freezing direction,
respectively [36].
Silk fibroin gels are important materials due to their attractive
properties such as a high mechanical strength, biocompatibility,
and controlled degradability [41e43]. Gelation of silk fibroin in
aqueous solutions mainly occurs by self-assembly of fibroin mole-
cules via intermolecular b-sheet crystallites acting as physical
cross-links. The formation of b-sheets and hence fibroin gelation
can be induced by several triggers such as pH [44], temperature
[45], fibroin concentration [46], cations [47], diepoxide cross-
linkers [48], vortexing [49], and electrical field [50]. In tissue en-
gineering applications, a high mechanical strength and an inter-
connected open pore structure with micrometer-sized pores are
essential considerations in the development of silk fibroin gels and
scaffolds. In addition, an anisotropic microstructure and anisotropic
mechanical properties are also required in fibroin scaffolds to
mimic the biological tissues. To our knowledge, there is only one
report on the preparation of silk fibroin scaffolds with anisotropic
mechanical properties [30]. These scaffolds were prepared by
uniaxial freezing of aqueous fibroin solutions followed by freeze-
drying to fix the anisotropic microstructure formed due to the ice
template. Because the scaffolds are soluble in aqueous environ-
ment, they were finally treated with methanol to induce the for-
mation of b-sheets [30]. The scaffolds thus produced exhibit a low
mechanical strength, e.g., their Young's modulus is below 4 kPa
[30]. This is expected due to the fact that the cross-linking occurs
after forming the pore walls of the scaffold leading to the formation
of weak intermolecular bonds.
The aim of this study was to prepare high-strength biocom-
patible scaffolds with anisotropic mechanical properties. Here, we
describe preparation of anisotropic silk fibroin cryogels and scaf-
folds exhibiting a Young's modulus in the range of MPa that sustain
up to 20 MPa compressive stresses. The cryogels were prepared by a
combined directional freezing e cryogelation process starting from
an aqueous 4.2 wt% fibroin solution containing butanediol digly-
cidyl ether cross-linker and N,N,N0 ,N0 -tetramethylethylenediamine
(TEMED). In the first step, the reactor containing the aqueous so-
lution of fibroin, cross-linker, and TEMED was immersed into liquid
nitrogen at a controlled rate to create a directionally frozen ice
template. In the second step, the cryogelation reactions were con-
ducted in this frozen solution at 18  C whereby the cryo-
concentrated fibroin in the unfrozen microzones of the reaction
system forms a 3D fibroin network. The scaffolds exhibit aniso-
tropic microstructure and hence anisotropic mechanical properties,
e.g., the Young's modulus is 3.4 ± 0.5 MPa and 0.8 ± 0.3 MPa when
measured along the directions parallel and vertical to the freezing
direction, respectively. As will be seen below, all the cryogels could
completely be compressed due to squeezing out of water from their
pores. Upon removal of the load, the compressed cryogels imme-
diately recover their original dimensions and mechanical proper-
ties by absorbing the released water into their pores.
2. Experimental section
2.1. Materials
The cross-linker butanediol diglycidyl ether (BDDE, Sigma-
Aldrich), N,N,N0 ,N0 -tetramethylethylenediamine (TEMED, Sigma-
Aldrich), Na2CO3 (Merck), and LiBr (Merck) were used as received.
Bombyx mori cocoons were purchased from Bursa Association of
Agricultural Sales Cooperatives for Silk Cocoons (Kozabirlik,
Turkey). Silk fibroin was separated from cocoons by boiling them
for 1 h in aqueous solution of 0.02 M Na2CO3 to remove the sericin
proteins followed by washing the remaining fibroin three times
with distilled water at 70  C, for 20 min each [44]. Silk fibroin was
dissolved in aqueous 9.3 M LiBr at 60  C for ~2 h and then dialyzed
using dialysis tubing (10000 MWCO, Snake Skin, Pierce) for 3 days
against water that was changed three times a day [44]. After
centrifugation, the final concentration of silk fibroin in aqueous
solution was about 5 w/w%, which was determined by weighing the
remaining solid after drying.
2.2. Preparation of fibroin cryogels
Anisotropic fibroin cryogels were prepared by directional
freezing of aqueous 4.2 wt% fibroin solution containing BDDE cross-
linker and TEMED (0.25 v/v%) in liquid nitrogen followed by con-
ducting the cryogelation reactions at 18  C for 24 h. BDDE con-
centration in the fibroin solution was set to 20 mmol epoxy per
gram of fibroin [48]. Typically, 5 mL of 5 wt% fibroin solution were
mixed with BDDE (0.50 mL), TEMED (15 mL), and water to make the
final volume 6 mL. The homogeneous fibroin solution was then
transferred into several 1 mL plastic syringes of 4 mm internal
diameter. Each plastic syringe containing the gelation solution was
then connected to the upper clamp of the Zwick-Roell test machine,
which was moved downward at a constant rate R into a cold bath
containing liquid nitrogen. Experimental apparatus for directional
freezing of aqueous 4.2 wt% fibroin containing BDDE and TEMED is
shown in Fig. 1a. The immersion rate R of the syringes into liquid
nitrogen controlled by the software of the test machine was varied
between 2.5 and 35 mm min1. The syringes were then placed in a
cryostat at 18  C to conduct the cryogelation reactions for 1 day.
Control experiments were also carried out as described above,
except that the directional freezing step was not applied to the
fibroin solutions.
 B. Yetiskin, O. Okay / Polymer 112 (2017) 61e70 63

Fig. 1. (A): Experimental apparatus for the directional freezing of aqueous fibroin solutions. The syringes containing aqueous solution of fibroin, BDDE and TEMED are immersed in
liquid nitrogen at a controlled rate R before conducting the cryogelation reactions at 18  C. (B): Amide I region of ATR-FTIR spectra of freeze-dried silk fibroin before (dashed
curve), and after directional freezing - cryogelation (solid curves) at R ¼ 2.5 (dark red) and 35 mm min1 (dark blue). (C): Typical stress-strain curves of cryogel samples up to around
99% compressions where the nominal stress snom and true stress strue are plotted against the compressive strain ε. R ¼ 20 mm min1. The samples were compressed parallel to the
freezing direction. The blue arrows indicate calculations of the fracture nominal stress sf and fracture strain εf from the maximum in strue - l plot. (For interpretation of the
references to colour in this figure legend, the reader is referred to the web version of this article.)

2.3. Swelling and gel fraction measurements
The frozen cryogel samples after preparation were first thawed
at room temperature for 30 min and then immersed in a large
excess of water for at least 3 days by replacing water several times
to extract any soluble species. The swelling equilibrium was tested
by weighing the gel samples as well as by measuring the diameter
of the gel samples using a calibrated digital compass (Mitutoyo
Digimatic Caliper, Series 500, resolution: 0.01 mm). The equilib-
rium swollen gel samples were taken out of water and immediately
frozen at 25  C for 1 day. The samples were freeze-dried at 40  C
under 0.12 mbar vacuum for 1 day and at 60  C under 0.011 mbar
for an additional 1 day. The freeze-drying system consisted of a
freeze-dryer (Christ Alpha 2e4 LD plus) connected to a vacuum
pump (Vacuubrand RZ 6). The pressure during freeze-drying was
adjusted using a valve controller and monitored by an active digital
controller. All cryogel samples were freeze-dried under the same
conditions. The equilibrium weight qw and volume swelling ratios
qv were calculated as qw ¼ m=mdry and qv ¼ ðD=Ddry Þ3 , respec-
tively, where m and D are the mass and the diameter of the equi-
librium swollen gel sample, respectively, and mdry and Ddry are the
same quantities for the freeze-dried sample. The gel fraction Wg,
that is, the conversion of fibroin to the 3D fibroin network (mass of
water-insoluble fibroin/initial mass of the fibroin in the feed) was
calculated from the masses of dry fibroin network and from the
fibroin in the feed.
2.4. ATR-FTIR measurements
Fourier-transform infrared (FTIR) spectra of the freeze-dried
samples were obtained using a single bounce diamond attenu-
ated total refractance (ATR) module on a FTIR spectrometer (Nicolet
Nexus 6700) equipped with a liquid nitrogen cooled mercury-
cadmium-telluride (MCT) detector. The resolution of each spec-
trum was 4 cm1, and 64 interferograms were coadded in the range
of 500e4000 cm1. The dashed and solid curves in Fig. 1b present
the Amide I band region (1580e1720 cm1) of FTIR spectra of
freeze-dried fibroin before and after directional freezing - cry-
ogelation, respectively. The peak at 1640 cm1 corresponding to the
random coil conformation disappears after cryogelation and a new
peak at 1620 cm1 appears which was assigned to b-sheet
conformation [51,52]. In addition to this peak, shoulders at 1660
and 1698 cm1 are seen after cryogelation, which were assigned to
the a-helix and b-turn conformations, respectively. To estimate the
conformation of fibroin, FTIR spectra of the samples were analyzed
using PeakFit software (Version 4.12, SeaSolve Software Inc.). Linear
baseline correction was applied to the Amide I region before the
band was deconvolved by Gauss Amplitude function [48]. For the
curve fitting procedure, the initial band positions at 1620, 1640,
1660, and 1698 cm1, representing b-sheet, random coil, a-helix,
and b-turn conformations, respectively, were fixed, allowing their
widths and heights to vary (Fig. S1).
2.5. Mechanical tests
Uniaxial compression measurements were conducted on cry-
ogel samples both in equilibrium swollen and dried states. The
samples were cut from the directions parallel and vertical to the
freezing direction to obtain a rectangular shape of dimensions
33 ± 6 mm  32 ± 4 mm  2.2 ± 0.4 mm. The compression tests
were performed at 23 ± 2  C on a Zwick Roell test machine using a
500 N load cell [53]. An initial compressive contact to 0.05 N was
applied to ensure a complete contact between the sample and the
surface. Load and displacement data were collected during the
experiments at a constant crosshead speed of 0.3 mm min1.
Compressive stress was presented by its nominal snom and true
values strue, which are the force per cross-sectional area of the
undeformed and deformed specimen, respectively. Assuming the
sample volume remains constant during deformation, the true
stress strue was calculated as strue ¼ l snom where l is the defor-
mation ratio (deformed length/original length). The compressive
strain is given by the compression ratio ε which is the change in the
sample length relative to its initial length, i.e., ε ¼ 1 - l. Fig. 1c shows
typical stress-strain curves of fibroin cryogels, where the nominal
64 B. Yetiskin, O. Okay / Polymer 112 (2017) 61e70
snom and true stresses strue are plotted against the strain ε. It is seen
that although snom increases continuously with increasing strain up
to about 99% compression, the corresponding strue - l plot passes
through a maximum revealing the onset of a failure in the cryogel
specimen [53]. Therefore, the fracture nominal stress sf and the
fracture strain εf were calculated from the maxima in strue - l plots,
as indicated by the blue arrows in the figure. The compression
modulus E was calculated from the slope of stress-strain curves
between 2 and 4% compressions while the stress at 3% compression
was reported as the compressive stress scomp. For reproducibility, at
least six samples were measured for each sample and the results
were averaged.
2.6. Texture determination
Cross-sectional scanning electron microscopy (SEM) studies
were performed on freeze-dried cryogel samples at various mag-
nifications between 20 and 1000 times in a field emission scanning
electron microscope (JEOL JSM-6510LV). All samples were coated
with gold for 3 min using a Sputter coater (S150 B Edwards) prior to
the measurements. Texture determination of the freeze-dried and
swollen cryogel samples was also carried out using an image
analyzing system consisting of a microscope (XSZ single Zoom
microscope), a CDD digital camera (TK 1381EG) and a PC with the
data analyzing system Image-Pro Plus. For this purpose, cylindrical
gel samples of 3.7 mm in diameter were cut into thin slices of about
1 mm in thickness. The measurements were conducted at magni-
fications between 10 and 100 times.
3. Results and discussion
Anisotropic silk fibroin scaffolds were prepared by lowering the
reactor containing an aqueous solution of fibroin, BDDE cross-
linker, and TEMED into liquid nitrogen at a controlled rate fol-
lowed by conducting the cryogelation reactions at 18  C for 24 h
(Fig. 1a). The immersion rate R of the gelation solution into liquid
nitrogen was varied between 2.5 and 35 mm min1. In control
experiments, cryogelation reactions were conducted without the
Fig. 2. (A) The gel fraction Wg (open symbols), equilibrium weight qw (filled circles) and volume swelling ratios qv of the cryogels (filled triangles) shown as a function of the
immersion rate R. The data at zero immersion rate are for isotropic cryogels. (B) Spacing between fibroin layers, i.e., channel width in anisotropic cryogel scaffolds shown as a
function of R.
immersion period, i.e., by cooling the solution from 23 ± 2 to 18  C
and keeping at this temperature for 24 h to obtain isotropic
cryogels.
Fig. 2a shows the equilibrium weight qw and volume swelling
ratios qv of the cryogels and the gel fraction Wg plotted against the
immersion rate R. The data for isotropic cryogels are shown at R ¼ 0,
as indicated by an arrow. The gel fraction Wg representing the
conversion of water-soluble fibroin to a 3D fibroin network is
1.3 ± 0.2 indicating a complete conversion, and the presence of
bound water that cannot be separated by freeze-drying [54]. As
detailed before [48,55], BDDE reacts with the amino groups on
fibroin to form intermolecular cross-links, whereby the mobility of
fibroin molecules significantly reduces triggering the conforma-
tional transition in fibroin from random coil to b-sheet structure.
The insolubility of fibroin after the directional freezing - cry-
ogelation process also suggests formation of b-sheet crystallites
acting as interstrand cross-links. Indeed, the results of b-sheet
contents estimated from the analysis of the Amide I band region of
FTIR spectra show that fibroin chains before gelation have 12 ± 2%
b-sheet structures, while their contribution increases to 28 ± 1 and
31 ± 2% in isotropic and anisotropic cryogels, respectively (Fig. S2).
The higher b-sheet content of anisotropic cryogels as compared to
the isotropic ones indicates that freezing in liquid nitrogen before
cryogelation promotes b-sheet crystallization. Fig. 2a also reveals
that both the weight qw and volume swelling ratio qv of anisotropic
cryogels are independent on R and they equal to 12.3 ± 0.4 and
1.2 ± 0.1, respectively. Isotropic cryogels exhibit a slightly higher
degree of swelling and a gel fraction Wg of 1.1 ± 0.1 indicating that
the immersion period increases the amount of bound water in the
cryogels.
The fact that the mass of the cryogels 12-fold increases upon
swelling in water while their volume remains almost constant
suggests that the swelling process of the cryogels is mainly gov-
erned by filling of their pores with water [56]. Assuming the pore
volume remains constant during swelling, the total porosity P can
be estimated from the weight and volume swelling ratios using the
equation [56]: P ¼ 1  qv ½1 þ ðqw  1Þ r1 where r is the fibroin
density (1.35 g/mL) [57]. Calculations show that the porosity P of
 B. Yetiskin, O. Okay / Polymer 112 (2017) 61e70
both isotropic and anisotropic cryogels is 93 ± 1% indicating their
highly porous structure. Indeed, cross-sectional optical images of
isotropic and anisotropic cryogel samples shown in Fig. 3a and
3bed, respectively, reveal existence of micrometer-sized pores. The
isotropic cryogel has a disordered pore structure within an inter-
connected 3D fibroin network while all cryogels prepared after
directional freezing e cryogelation process possess long aligned
fibroin layers interconnected with fibroin branches forming a 3D
network containing macropores. The images in Fig. 3bed also show
that increasing immersion rate R of the reaction solution into liquid
nitrogen prior cryogelation decreases the spacing between the
fibroin layers, i.e., the width of microchannels.
Fig. 4a and b shows typical scanning electron micrographs of
cross-sections of isotropic (a) and anisotropic cryogel scaffolds
formed at R ¼ 30 mm min1 (b). Isotropic cryogel sample consists of
randomly oriented spherical and ovaloid pores of 30 ± 11 mm in
diameter while the anisotropic sample possesses several hundred
micrometers long, aligned fibroin layers producing a channel-like
porous structure. The fibroin layers are interconnected by
perpendicularly oriented branches to form a “fishbone”
morphology. We have to mention that all the SEM and optical
images presented here are cross-sectional views of the scaffolds
and therefore, the aligned layers and channels are perpendicularly
oriented to the immersion direction. For instance, Fig. 4c and
d showing SEM images of the cryogels formed at R ¼ 20 and
25 mm min1, respectively, at a low magnification reveals radially
oriented pore structure of the scaffolds. These results suggest that
the growth of ice crystals occurs from the surface of the syringes
which is in conduct with the cooling liquid at 196  C to its interior
rather than in the direction in which the syringe is lowered into
liquid nitrogen. A similar observation was recently reported by
Arrua and Hilder in the preparation of cross-linked polyacrylates by
directional freezing and photoinitiated cryocopolymerization [39].
We attribute the directional freezing in radial direction to the
higher effective thermal conductivity in radial direction as
compared to the longitudinal (immersion) direction because the
heat flux must travel a shorter path in the former direction.
Fig. 5 shows a scheme representing the unidirectional growth of
ice crystals and formation of fibroin layers interconnected by
branches. The temperature gradient in the fibroin solution induced
by the contact with liquid nitrogen results in growing of ice from
the surface of the cylindrical reactor toward the center. As water
freezes, fibroin molecules together with BDDE cross-linker and
Fig. 3. Optical images of isotropic (a) and anisotropic fibroin cryogels (bed). R ¼ 10 (b), 25 (c), and 35 mm min1. Scaling bars in 3aed are 1 mm (left) and 100 mm (right).
65
TEMED are expelled from the formed ice crystals and accumulate
around the growing crystals to form unfrozen domains of the re-
action system. Thus, in addition of the temperature gradient, a
concentration gradient is established across the moving freezing
front due to the cryo-concentration of the reactants. This causes
Mullins-Sekerka instability leading to the collapse of the moving
freezing front leaving behind pockets of unfrozen concentrated
fibroin solution [7,8,58]. After the cryogelation reactions at 18  C,
the cryo-concentrated unfrozen fibroin solution between the
aligned ice crystals turns to a dense gel making the fibroin layers
and branches while ice act as template to form pores after thawing
and freeze-drying. The primary periodicity of the aligned porous
structures is defined by the Mulling-Sekerka instability wavelength
[58]. This wavelength is determined by competition between the
destabilizing solute interfacial concentration gradient and the sta-
bilizing effects acting to preserve the planarity of the interface, e.g.,
the imposed temperature gradient and surface energy of the solid/
liquid interface [58]. The layer spacing, i.e., the channel diameter of
14 ± 7 mm in Fig. 4b well compares with the instability wavelength
of 15 mm for aqueous PVA solutions [8].
The results also show that the spacing between the fibroin
layers strongly depends on the immersion rate R. Fig. 6 showing
SEM images of fibroin scaffolds formed at various R between 2.5
and 35 mm min1 indicate that the layer spacing is inversely pro-
portional to the immersion rate. The average spacing between the
layers were calculated by analyzing SEM images at various mag-
nifications. The results are collected in Fig. 2b where the layer
spacing is plotted against R. At low immersion rates R, the layer
spacing approaches to 70 mm with a broad length distribution while
at R > 20 mm min1, it is rather uniform and around 15 mm.
Decreasing layer spacing with increasing immersion rate is due to
the simultaneous increase of the freezing rate leading to the for-
mation of a larger number of growing ice crystals [32]. Some in-
ternal microcracks were also observed in the cryogel samples
formed at high immersion rates (25 mm min1) which we attri-
bute to rapid cooling of the fibroin solution (Fig. S3).
Anisotropic microstructure of the cryogels resulted in aniso-
tropic mechanical properties, as demonstrated by uniaxial
compression tests. The tests were performed by cutting dry and
swollen cryogel specimens from the directions parallel and vertical
to the freezing direction and compressing them at a constant speed
up to complete compression. Fig. 7 shows stress-strain curves of the
cryogel scaffolds formed at various R where the nominal stress snom
66 B. Yetiskin, O. Okay / Polymer 112 (2017) 61e70

Fig. 4. SEM images of isotropic (a) and anisotropic cryogel scaffolds (bed). Scaling bars are100 mm (a, b) and 500 mm (c, d). R ¼ 20 (c), 25 (d), and 30 mm min1 (b).

Fig. 5. Cartoon representing formation of aligned porous structure in fibroin cryogels under directional freezing. Fibroin solution before (a) and after directional freezing (b), and
after cryogelation (c) is presented.

is plotted against the deformation ε. Note that the curves were
corrected by considering the onset of microcracks in true stress strue
e ε plots as detailed in the experimental section and in Fig. 1c. Solid
and dashed curves represent the results of the measurements along
the directions parallel and vertical to the freezing direction,
respectively. The insets show the portion of the curves below
snom ¼ 0.6 MPa. The cryogels sustain up to ~90% compression at
~20 MPa compressive stresses. This extraordinary mechanical
strength is due to the cryo-concentration of fibroin producing
dense pore walls of high fibroin concentration [32,33]. For instance,
freezing of an aqueous 6 wt% fibroin at 18  C results in a frozen
system composed of 88% ice and the rest being a concentrated
unfrozen solution containing 37 wt% fibroin, which is about 6-fold
larger than the nominal fibroin concentration [55]. Fig. 7 also shows
that, at below R ¼ 30 mm min1, the initial slope of stress-strain
curves corresponding to the Young's modulus E is higher in paral-
lel direction as compared to the vertical direction. Another point is
the appearance of a near-plateau regime at around 5% compression
in parallel direction during which the scaffolds easily deform under
force. As reported before [55], the appearance of a plateau regime
indicates that the porous structure is mechanically stable up to the
onset of the plateau while it starts to collapse with the onset of the
plateau. The results reveal higher mechanical stability of 3D porous
structure during compression in parallel direction as compared
vertical direction.
The mechanical properties of the cryogels are compiled in Fig. 8
where the Young's modulus E, compressive scomp and fracture
stresses sf of dry (left panel) and swollen cryogels (right panel) are
 B. Yetiskin, O. Okay / Polymer 112 (2017) 61e70
Fig. 6. SEM images of anisotropic cryogel scaffolds. R ¼ 2.5 (a), 10 (b), 20 (c), 25 (d), 30 (e), and 35 mm min1 (f). Scaling bars ¼ 100 mm (Magnification ¼ 100).
Fig. 7. Stress e strain curves of fibroin scaffolds as the dependence of the nominal
stress snom on the compressive strain ε. The measurements were conducted parallel
(solid curves) and vertical to the freezing direction (dashed curves). Immersion
rate ¼ 10 (a), 15 (b), 20 (c), and 30 mm min1 (d).
plotted against the immersion rate R. The circles and triangles
represent the data obtained from parallel and vertical directions,
respectively. Moreover, the data of isotropic cryogels in both di-
rections shown at R ¼ 0 are presented by the filled and open
squares. For both dry and swollen cryogel samples, the extent of
anisotropy first increases with increasing immersion rate R up to
20 mm min1 and then decreases again with a further increase in R.
67
The modulus E of scaffolds in parallel direction attains a maximum
value of 3.4 ± 0.5 MPa at R ¼ 20 mm min1, which is about four-fold
larger than that measured in perpendicular direction
(0.8 ± 0.3 MPa). Considering the Young's modulus of human cornea
is around 3 and 1 MPa in horizontal and vertical directions,
respectively [6], the results show that the cryogel scaffolds formed
at 20 mm min1 possess a comparable modulus.
The results also show that the anisotropic pore structure of the
scaffolds mainly affect the initial part of the stress-strain curves
represented by the modulus E and the compressive stress scomp at
3% compression while the ultimate properties such as the fracture
stress sf and fracture strain remain unaffected. This could be
explained by considering the microstructure of anisotropic cryogels
(Figs. 3, 4 and 6): Compressing in parallel to the freezing direction
means vertically compressing several hundred nanometers long,
radially oriented fibroin layers while, in vertical direction, the
branches between the layers are compressed, which are only tens
of nanometers. Thus, a much larger stress is initially required to
compress the fibroin layers as compared to their branches. As the
strain is increased, the porous structure starts to collapse resulting
in a decrease of the channel width so that the stress required to
deform the cryogel becomes independent of direction. The results
also show that all anisotropic scaffolds formed at R  20 mm min1
sustain at least 20 MPa compressive stresses as compared to
8 ± 1 MPa for the isotropic scaffold (Fig. 8). The results of swollen
cryogels shown in the right panel of Fig. 8 reveals that the modulus,
compressive and fracture stresses of the cryogels decrease by one
order of magnitude after swelling in water.
The key advantage of the present directional freezing - cry-
ogelation process is the shape and mechanical property recover-
ability of anisotropic cryogels subjected to complete compressions.
This is presented in Fig. 9a and b showing successive compression
test results conducted up to about 99% compression. Here, the solid
curves show stress-strain curves of cryogel samples prepared at
R ¼ 20 mm min1 as the dependences of the nominal snom and true
stresses strue on the compressive strain ε, respectively. The cryogel
samples were compressed parallel (left panel) and vertical to the
68 B. Yetiskin, O. Okay / Polymer 112 (2017) 61e70

Fig. 8. The compressive modulus E, compressive scomp and fracture stresses sf of dry (left panel) and swollen cryogels (right panel) plotted against the immersion rate. The circles
and triangles represent the data obtained from parallel and perpendicular directions, respectively, while the data of isotropic cryogels in both directions are shown by the filled and
open squares.

freezing direction (right panel). It is seen that, although the cryogel
sample could completely be compressed, a maximum in strue vs ε
plot appears at 95% compression suggesting the formation of
microcracks in the sample. After this compression test, the cryogel
sample remained in a compressed state due to squeezing out of
water from its pores, as seen in the images in Fig. 9c (c1/c3). After
release of the piston, the compressed sample absorbs the released
water (c3/c5) while adding water completely recovers its original
dimension by sucking up water in its pores (c6), so that it can be
subjected to the next compression test (see also: Supporting
Information Movie). Dashed and dash-dot curves in Fig. 9a and b
represent 2nd to 4th compression test results, each conducted after
adding water to the gel sample. The insets to the figures reveal a
good superposition of the successive stress-strain curves demon-
strating that the damage in the cryogel is self-healed upon
removing the stress. The average compression at break and fracture
stress of four successive tests are 94 ± 1% and 8.4 ± 1.4 MPa,
respectively. Similar results were also obtained for isotropic and
anisotropic cryogels formed at various immersion rates. Thus, all
fibroin cryogels can be compressed up to about 99% strain without
any permanent failure.
Supplementary data related to this article can be found online at
http://dx.doi.org/10.1016/j.polymer.2017.01.079.
4. Conclusions
Silk fibroin cryogels and scaffolds with anisotropic microstruc-
tures and anisotropic mechanical properties could be obtained by a
combined directional freezing e cryogelation technique starting
from aqueous 4.2 wt% fibroin solutions containing BDDE cross-
linker and TEMED. In the first step, the reactor containing the
aqueous solution of fibroin, cross-linker, and TEMED was immersed
into liquid nitrogen at a controlled rate to create a directionally
frozen ice template. In the second step, the cryogelation reactions
were conducted in this frozen solution at 18  C whereby the cryo-
concentrated fibroin in the unfrozen microzones of the reaction
system forms a 3D fibroin network. The scaffolds exhibit a Young's
modulus in the range of MPa and sustain up to 20 MPa compressive
stresses. In addition to high mechanical strength, they also exhibit
anisotropic microstructure and hence anisotropic mechanical
properties, e.g., the Young's modulus is 3.4 ± 0.5 MPa and
0.8 ± 0.3 MPa when measured along the directions parallel and
 B. Yetiskin, O. Okay / Polymer 112 (2017) 61e70 69

Fig. 9. (A, B) Four successive compression test results conducted on anisotropic cryogel samples up to around 99% compression where snom (a) and strue (b) are plotted against the
strain ε. The gel samples were prepared parallel (left) and perpendicular to the freezing direction (right). R ¼ 20 mm min1. (C) Images showing the initial dimension of the cryogel
sample (c1), its compressed state (c2), and the recovery of the initial dimension after release of the piston (c2/c5), and after adding water (c6).

vertical to the freezing direction, respectively. All the cryogels could
completely be compressed due to squeezing out of water from their
pores. Upon removal of the load, the compressed cryogels imme-
diately recover their original dimensions and mechanical proper-
ties by absorbing the released water into their pores. Because many
biological tissues possess anisotropic hierarchical morphologies,
high-strength anisotropic fibroin cryogels and scaffolds presented
here have a broad range of potential applications in tissue engi-
neering, bioseparation, microfluidics, and organic electronics.
Acknowledgments
This work was supported by the Scientific and Technical
Research Council of Turkey (TUBITAK), KBAG 114Z312. OO thanks
the Turkish Academy of Sciences for the partial support.
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.polymer.2017.01.079.
References
[1] O. Okay, General properties of hydrogels, in: G. Gerlach, K.F. Arndt (Eds.),
Hydrogel Sensors and Actuators. Springer Series on Chemical Sensors and
Biosensors vol. 6, Springer Berlin Heidelberg, 2009, pp. 1e14.
[2] A.C. Ahn, T.J. Kaptchuk, Spatial anisotropy analyses of subcutaneous tissue
layer: potential insights into its biomechanical characteristics, J. Anat. 219
(2011) 515e524.
[3] L.N. Williams, S.H. Elder, J.L. Bouvard, M.F. Horstemeyer, The anisotropic
compressive mechanical properties of the rabbit patellar tendon, Biorheology
45 (2008) 577e586.
[4] J.S. Jurvelin, M.D. Buschmann, E.B. Hunziker, Mechanical anisotropy of the
human knee articular cartilage in compression, Proc. Inst. Mech. Eng. H. 217
(2003) 215e219.
[5] N.L. Nerurkar, D.M. Elliott, R.L. Mauck, Mechanical design criteria for inter-
vertebral disc tissue engineering, J. Biomech. 43 (2010) 1017e1030.
[6] A.C. Jayasuriya, S. Ghosh, J.I. Scheinbeim, V. Lubkin, G. Bennett, P. Kramer,
A study of piezoelectric and mechanical anisotropies of the human cornea,
Biosens. Bioelectron. 18 (2003) 381e387.
[7] H. Zhang, A.I. Cooper, Aligned porous structures by directional freezing, Adv.
Mater 19 (2007) 1529e1533.
[8] H. Zhang, I. Hussain, M. Brust, M.F. Butler, S.P. Rannard, A.I. Cooper, Aligned
two-and three-dimensional structures by directional freezing of polymers and
nanoparticles, Nat. Mater 4 (2005) 787e793.
[9] L.E. Millon, H. Mohammadi, W.K. Wan, Anisotropic polyvinyl alcohol hydrogel
for cardiovascular applications, J. Biomed. Mat. Res. B Appl. Biomater. 79
(2006) 305e311.
[10] W. Wan, A.D. Bannerman, L. Yang, H. Mak, Poly (vinyl alcohol) cryogels for
biomedical applications, Adv. Polym. Sci. 263 (2014) 283e321.
[11] S. Chatelin, M. Bernal, T. Deffieux, C. Papadacci, P. Flaud, A. Nahas, C. Boccara,
70 B. Yetiskin, O. Okay / Polymer 112 (2017) 61e70
J.L. Gennisson, M. Tanter, M. Pernot, Anisotropic polyvinyl alcohol hydrogel
phantom for shear wave elastography in fibrous biological soft tissue: a
multimodality characterization, Phys. Med. Biol. 59 (2014) 6923e6940.
[12] C.M. Voge, M. Kariolis, R.A. MacDonald, J.P. Stegemann, Directional conduc-
tivity in SWNT-collagen-fibrin composite biomaterials through strain-induced
matrix alignment, J. Biomed. Mat. Res. A 86 (2008) 269e277.
[13] A. Osorio-Madrazo, M. Eder, M. Rueggeberg, J.K. Pandey, M.J. Harrington,
Y. Nishiyama, J.L. Putaux, C. Rochas, I. Burgert, Reorientation of cellulose
nanowhiskers in agarose hydrogels under tensile loading, Biomacromolecules
13 (2012) 850e856.
[14] M.C. Swan, D.G. Bucknall, T.E. Goodacre, J.T. Czernuszka, Synthesis and
properties of a novel anisotropic self-inflating hydrogel tissue expander, Acta
Biomater. 7 (2011) 1126e1132.
[15] W. Yang, H. Furukawa, J.P. Gong, Highly extensible double-network gels with
self-assembling anisotropic structure, Adv. Mater 20 (2008) 4499e4503.
[16] M. Haque, G. Kamita, T. Kurokawa, K. Tsujii, J.P. Gong, Unidirectional align-
ment of lamellar bilayer in hydrogel: one-dimensional swelling, anisotropic
modulus, and stress/strain tunable structural color, Adv. Mater 22 (2010)
5110e5114.
[17] X. Pei, T. Zan, H. Li, Y. Chen, L. Shi, Z. Zhang, Pure anisotropic hydrogel with an
inherent chiral internal structure based on the chiral nematic liquid crystal
phase of rodlike viruses, ACS Macro Lett. 4 (2015) 1215e1219.
[18] S. Ahadian, J. Ramo
n-Azco

n, M. Estili, X. Liang, S. Ostrovidov, H. Shiku,
M. Ramalingam, K. Nakajima, Y. Sakka, H. Bae, T. Matsue, Hybrid hydrogels
containing vertically aligned carbon nanotubes with anisotropic electrical
conductivity for muscle myofiber fabrication, Sci. Rep. 4 (2014) 4271.
[19] K.J. Walker, S.V. Madihally, Anisotropic temperature sensitive chitosan-based
injectable hydrogels mimicking cartilage matrix, J. Biomed. Mat. Res. B Appl.
Biomater. 103 (2015) 1149e1160.
[20] L.A. Hockaday, K.H. Kang, N.W. Colangelo, P.Y. Cheung, B. Duan, E. Malone,
J. Wu, L.N. Girardi, L.J. Bonassar, H. Lipson, C.C. Chu, Rapid 3D printing of
anatomically accurate and mechanically heterogeneous aortic valve hydrogel
scaffolds, Biofabrication 4 (2012) 035005.
[21] L. Zhang, J. Zhao, J. Zhu, C. He, H. Wang, Anisotropic tough poly (vinyl alcohol)
hydrogels, Soft Matter 8 (2012) 10439e10447.
[22] H.M. Tong, I. Noda, C.C. Gryte, Formation of anisotropic ice-agar composites by
directional freezing, Colloid Polym. Sci. 262 (1984) 589e595.
[23] F. Yokoyama, E.C. Achife, J. Momoda, K. Shimamura, K. Monobe, Morphology
of optically anisotropic agarose hydrogel prepared by directional freezing,
Colloid Polym. Sci. 268 (1990) 552e558.
[24] N.W. Meghri, A.E. Donius, B.W. Riblett, E.J. Martin, A.M. Clyne, U.G.K. Wegst,
Directionally solidified biopolymer scaffolds: mechanical properties and
endothelial cell responses, JOM 62 (2010) 71e75.
[25] X. Qi, J. Ye, Y. Wang, Alginate/poly (lactic-co-glycolic acid)/calcium phosphate
cement scaffold with oriented pore structure for bone tissue engineering,
J. Biomed. Mater Res. A 89 (2009) 980e987.
[26] N.L. Francis, P.M. Hunger, A.E. Donius, B.W. Riblett, A. Zavaliangos, U.G. Wegst,
M.A. Wheatley, An ice-templated, linearly aligned chitosan-alginate scaffold
for neural tissue engineering, J. Biomed. Mater Res. A 101 (2013) 3493e3503.
[27] S.R. Caliari, B.A. Harley, The effect of anisotropic collagen-GAG scaffolds and
growth factor supplementation on tendon cell recruitment, alignment, and
metabolic activity, Biomaterials 32 (2011) 5330e5340.
[28] X. Wu, Y. Liu, X. Li, P. Wen, Y. Zhang, Y. Long, X. Wang, Y. Guo, F. Xing, J. Gao,
Preparation of aligned porous gelatin scaffolds by unidirectional freeze-drying
method, Acta Biomater. 6 (2010) 1167e1177.
[29] J. Guan, D. Porter, K. Tian, Z. Shao, X. Chen, Morphology and mechanical
properties of soy protein scaffolds made by directional freezing, J. Appl.
Polym. Sci. 118 (2010) 1658e1665.
[30] M.C. Asuncion, J.C. Goh, S.L. Toh, Anisotropic silk fibroin/gelatin scaffolds from
unidirectional freezing, Mat. Sci. Eng. C Mat. Biol. Appl. 67 (2016) 646e656.
[31] Q. Zhang, Y. Zhao, S. Yan, Y. Yang, H. Zhao, M. Li, S. Lu, D.L. Kaplan, Preparation
of uniaxial multichannel silk fibroin scaffolds for guiding primary neurons,
Acta Biomater. 8 (2012) 2628e2638.
[32] O. Okay, V.I. Lozinsky, Synthesis and structureeproperty relationships of
cryogels, Adv. Polym. Sci. 263 (2014) 103e157.
[33] V.I. Lozinsky, O. Okay, Basic principles of cryotropic gelation, Adv. Polym. Sci.
263 (2014) 49e101.
[34] S. Dogu, O. Okay, Tough Organogels Based on Polyisobutylene with Aligned
Porous Structures Polymer, vol. 49, 2008, pp. 4626e4634.
[35] M. Chen, J. Zhu, G. Qi, C. He, H. Wang, Anisotropic hydrogels fabricated with
directional freezing and radiation-induced polymerization and crosslinking
method, Mat. Lett. 89 (2012) 104e107.
[36] J. Wu, Q. Zhao, J. Sun, Q. Zhou, Preparation of poly (ethylene glycol) aligned
porous cryogels using a unidirectional freezing technique, Soft Matter 8
(2012) 3620e3626.
[37] M. Barrow, A. Eltmimi, A. Ahmed, P. Myers, H. Zhang, Frozen polymerization
for aligned porous structures with enhanced mechanical stability, conduc-
tivity, and as stationary phase for HPLC, J. Mat. Chem. 22 (2012)
11615e11620.
[38] D. Zhao, J. Zhu, Z. Zhu, G. Song, H. Wang, Anisotropic hierarchical porous
hydrogels with unique water loss/absorption and mechanical properties, RSC
Adv. 4 (2014) 30308e30314.
[39] R.D. Arrua, E.F. Hilder, Highly ordered monolithic structures by directional
freezing and UV-initiated cryopolymerisation. Evaluation as stationary phases
in high performance liquid chromatography, RSC Adv. 5 (2015) 71131e71138.
[40] M. Chau, K.J. De France, B. Kopera, V.R. Machado, S. Rosenfeldt, L. Reyes,
K.J. Chan, S. Fo €rster, E.D. Cranston, T. Hoare, E. Kumacheva, Composite
hydrogels with tunable anisotropic morphologies and mechanical properties,
Chem. Mater 28 (2016) 3406e3415.
[41] C. Vepari, D.L. Kaplan, Silk as a biomaterial, Prog. Polym. Sci. 32 (2007)
991e1007.
[42] F. Vollrath, D. Porter, Silks as ancient models for modern polymers, Polymer
50 (2009) 5623e5632.
[43] J.G. Hardy, L.M. Romer, T.R. Scheibel, Polymeric materials based on silk pro-
teins, Polymer 49 (2008) 4309e4327.
[44] U.J. Kim, J. Park, C. Li, H.J. Jin, R. Valluzzi, D.L. Kaplan, Structure and properties
of silk hydrogels, Biomacromolecules 5 (2004) 786e792.
[45] A. Matsumoto, J. Chen, A.L. Collette, U.J. Kim, G.H. Altman, P. Cebe, D.L. Kaplan,
Mechanisms of silk fibroin sol-gel transitions, J. Phys. Chem. B 110 (2006)
21630e21638.
[46] H.J. Jin, D.L. Kaplan, Mechanism of silk processing in insects and spiders,
Nature 424 (2003) 1057e1061.
[47] C. Dicko, J.M. Kenney, D. Knight, F. Vollrath, Transition to a beta-sheet-rich
structure in spidroin in vitro: the effects of pH and cations, Biochemistry 43
(2004) 14080e14087.
[48] Karakutuk, F. Ak, O. Okay, Diepoxide-triggered conformational transition of
silk fibroin: formation of hydrogels, Biomacromolecules 13 (2012)
1122e1128.
[49] T. Yucel, P. Cebe, D.L. Kaplan, Vortex-induced injectable silk fibroin hydrogels,
Biophys. J. 97 (2009) 2044e2050.
[50] E. Servoli, D. Maniglio, A. Motta, C. Migliaresi, Folding and assembly of fibroin
driven by an AC electric field: effects on film properties, Macromol. Biosci. 8
(2008) 827e835.
[51] X. Chen, D.P. Knight, Z. Shao, F. Vollrath, Conformation transition in silk
protein films monitored by time-resolved Fourier transform infrared spec-
troscopy: effect of potassium ions on Nephila spidroin films, Biochemistry 41
(2002) 14944e14950.
[52] C. Mo, C. Holland, D. Porter, Z. Shao, F. Vollrath, Concentration state depen-
dence of the rheological and structural properties of reconstituted silk, Bio-
macromolecules 10 (2009) 2724e2728.
[53] A. Argun, V. Can, U. Altun, O. Okay, Non-ionic double and triple network
hydrogels of high mechanical strength, Macromolecules 47 (2014)
6430e6440.
[54] X. Hu, D. Kaplan, P. Cebe, Effect of water on the thermal properties of silk
fibroin, Thermochim. Acta 461 (2007) 137e144.
[55] F. Ak, Z. Oztoprak, I. Karakutuk, O. Okay, Macroporous silk fibroin cryogels,
Biomacromolecules 14 (2013) 719e727.
[56] O. Okay, Macroporous copolymer networks, Prog. Polym. Sci. 25 (2000)
711e779.
[57] S.J. Park, K.Y. Lee, W.S. Ha, S.Y. Park, Structural changes and their effect on
mechanical properties of silk fibroin/chitosan blends, J. Appl. Polym. Sci. 74
(1999) 2571e2575.
[58] M.F. Butler, Instability formation and directional dendritic growth of ice
studied by optical interferometry, Cryst. Growth Des. 1 (2001) 213e223.